Goto predicted that half of all new server chips shipped to hyperscalers in 2025 will use its architecture. “The number of enterprises using Arm-based chips for cloud processing has grown more than 14x since 2021, exceeding 70,000 in 2025,” he added.

This sounds like an extremely optimistic outlook.

I think the bigger story is that Intel is going from “blue chip” type tech stock to more like Seagate or WD (declining revenues on a historical basis).

Good win for Intel, but this is only got packaging, a foundry win would have been more impactful.

I am honestly thought Future PLC would honour their promise to keep the their historical article online, it can’t be that expensive to host them.

They are a decent “front-end” to google search.

They are also working on their indexing solution, but for French and German content.

Hah! I was actually thinking about that too.

Intel just can’t catch a break.

Lip-Bu Tan is not unique in his desire to keep selling to China. This is true of all American companies, technology companies or otherwise.

Investors seemed to interpret the potential tariff exemptions as a positive for Apple and other major tech companies that have been making huge financial commitments to manufacture more chips and other components in the U.S.

Big Tech already has made collective commitments to invest about $1.5 trillion in the U.S. since Trump moved back into the White House in January. That figure includes a $600 billion promise from Apple after the iPhone maker boosted its commitment by tacking another $100 billion on to a previous commitment made in February.

I do wonder how many of these commitments will come to pass and to what extent these are merely elements of a negotiation strategy.

I am good man, think whatever you want.

By all metrics (product performance, market share, capitalization/stock price) they are in free fall and have been for half a decade.

No need to be overly pedantic.

If Apple (or another one of the companies you listed) massively collapsed from their leadership position, it would also be a point of discussion around whether stock buyback was justified.

Mind you, I don’t think nationalisation is likely to help Intel or that it is a desirable outcome, I am just sharing the reasoning.

He is refering to Intel spending large sums of money on stock buybacks instead investing it in their business.

I got to give PCI AIG props for having such solid backwards and forwards compatibility.

That being said, it sounds like consumer level hardware will be on PCIe 5.0 for a long time.

PCIe 6.0 x4 SSDs for PCs, which would increase peak bandwidth to up to 32 GB/s, will not be available until 2030, according to Wallace C. Kuo, chief executive of Silicon Motion, a major independent developer of SSD controllers.

Source

I was reading about that earlier.

They really look their imploding.

So this is what the CCP authorities were referring to when bringing up backdoors in H20 enterprise GPUs a few weeks ago.

Hot take: the easiest way to massively increase compliance is to make Nvidia executives (Huang specifically) personally liable for export control violations. This is not silver bullet, but neither is any location tracking scheme. Nvidia will of course cry foul, but when personal criminal liability is on the line (and the US would have show that they are willing and able to implement it), I can guarantee you they’ll figure out a way to make it work. Necessity is the mother of invention as the saying goes.

Note that I am not saying this is a good or bad thing.

Investing another $400 billion in the US sounds like a bigger commitment than buying Intel.

I am not even sure what TSMC would do with 49% of Intel. Upgrade their Intel’s existing fabs using TSMC tech? I can’t imagine TSMC getting into developing semiconductors (e.g. x86 CPUs), doesn’t seem to fit their strategy and Intel only strong position is x86 CPUs for laptops.

A mere decrease of tariffs from 20% to 15% doesn’t seem all that enticing (without security guarantees for Taiwan), one would think TSMC could easily make their customers eat the 5% extra tariff cost, it’s not like they have much of an option.

Foxconn, which is also the world’s largest AI server maker, said the Lordstown facility will become an AI data center in the future and that Foxconn will play an active and important role in building its infrastructure.

“The purpose is to provide greater flexibility and revitalize assets in Lordstown to focus on the company’s AI data center business strategy, especially to help our clients to speed up the building of AI servers and AI data centers in the U.S.,” a Foxconn spokesperson told Nikkei Asia.

Foxconn is the world’s largest AI server maker? And they even operate data centres? This is news to me.

As of late last year, only around 5% of the Panther Lake chips that Intel printed were up to its specifications, these sources said. This yield figure rose to around 10% by this summer, said one of the sources, who cautioned that Intel could claim a higher number if it counted chips that did not hit every performance target. Reuters could not establish the precise yield at present.

10% yield sounds like a disastrous level.

I have zero clue about the nuances of yield improvement curves, but an increase from 5% to 10% in ~9 months does not sound good.

Sounds like Intel is going to be exiting the fab business.

Koduri’s startup appears to be focused on IP only, i.e. it’ll be down to another company to license the tech and make the chips themselves. That’s more cost-effective than having to purchase fabrication slots at the likes of TSMC, but as Arm will testify, processor IP licensing isn’t necessarily an easy path to big bucks.

Interesting choice of business model. They’ll still need a flagship product for their IP to have any value.

If it was possible to get this for the equivalent $250 (out of pocket price), this would be a pretty solid offering.

The cheapest 9060 XT (8 GB) is going for $385 where I live.